CN112102897A - Chemical vapor multicomponent deposition product component prediction method - Google Patents

Chemical vapor multicomponent deposition product component prediction method Download PDF

Info

Publication number
CN112102897A
CN112102897A CN202010988337.3A CN202010988337A CN112102897A CN 112102897 A CN112102897 A CN 112102897A CN 202010988337 A CN202010988337 A CN 202010988337A CN 112102897 A CN112102897 A CN 112102897A
Authority
CN
China
Prior art keywords
vibration
formula
gas
molecule
total
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010988337.3A
Other languages
Chinese (zh)
Other versions
CN112102897B (en
Inventor
关康
任海涛
曾庆丰
卢振亚
吴建青
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Huifengyuan Technology Co ltd
Original Assignee
South China University of Technology SCUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN202010988337.3A priority Critical patent/CN112102897B/en
Publication of CN112102897A publication Critical patent/CN112102897A/en
Application granted granted Critical
Publication of CN112102897B publication Critical patent/CN112102897B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C10/00—Computational theoretical chemistry, i.e. ICT specially adapted for theoretical aspects of quantum chemistry, molecular mechanics, molecular dynamics or the like
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Chemical & Material Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Medical Informatics (AREA)
  • Physiology (AREA)
  • Databases & Information Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Chemical Vapour Deposition (AREA)

Abstract

The invention discloses a method for predicting components of a chemical vapor multi-element deposition product, which comprises the following steps: step S1, calculating a distribution function according to the process conditions: step S2, calculating the hot melt and entropy; step S3, calculating standard enthalpy and Gibbs free energy according to the translation, rotation, vibration and electronic distribution function; step S4, obtaining the equilibrium output distribution of all products according to the chemical equilibrium principle, namely the mathematical condition of the minimum total Gibbs free energy of the system; step S5, the process conditions, the sedimented solid phase and the yield are used as input data, and a training model is established by using a BP algorithm. And step S6, after a BP training model is established, the calculated result is analyzed by combining a genetic algorithm, and the highest solid phase yield under different weight conditions is continuously searched out. The invention has the advantages that: and establishing a model capable of accurately predicting the components of the deposition theoretical product, realizing the optimization target of the maximum yield of the multiple components, accelerating the research and development efficiency of the CVD industrial production and reducing the production cost.

Description

Chemical vapor multicomponent deposition product component prediction method
Technical Field
The invention relates to the technical field of predicting deposition theoretical products, in particular to a chemical vapor multi-element deposition product component prediction method combining machine learning and multi-field coupling modeling.
Background
The ceramic matrix composite is a composite material which takes ceramic as a matrix and is compounded with other fibers. Has the good performances of high strength, high modulus, low density, high temperature resistance, wear resistance, corrosion resistance and the like. Especially the high temperature resistance of the ceramic matrix composite material, so that the application research of the ceramic matrix composite material in a high temperature environment is emphasized. However, the greatest defects of the ceramic material are that the ceramic material is brittle and is easily oxidized and corroded in a high-temperature water-oxygen environment. Therefore, the introduction of a coating or the design of a multi-component ceramic matrix on the surface of the ceramic matrix composite is an effective method for improving the high-temperature performance of the ceramic matrix composite.
Chemical Vapor Deposition (CVD) is the preferred method for preparing ceramic matrix composites by chemically reacting one or more Vapor compounds or elements containing the desired elements on the fiber surface to form a film or coating. Compared with other inorganic material preparation methods, the CVD method can prepare high-quality and high-purity coatings, and can realize the interface deposition of the components with complex shapes and control the distribution of components and substances through a process control group. However, the CVD process is very complicated, the intermediate gas phase products generated by the precursor reaction are very diverse, and there are many transition state species, and it is difficult to measure all the intermediate components by the existing experimental means.
In recent years, with the development of computer technology and quantum chemistry theory, quantum chemistry theory has been used for calculation
It has become possible to study the reaction mechanism and predict the direction of the reaction in depth, and there have been many successful paradigms in this regard.
The study of CVD reaction mechanisms using quantum chemical methods generally includes two categories, namely reaction thermodynamics and reaction kinetics. For a new CVD system, the reaction thermodynamics will generally be first calculated to predict the effect of process parameters (e.g., deposition temperature, total pressure, gas feed ratio, etc.) on the product produced at equilibrium conditions. Among them, the method based on the minimum Gibbs free energy of the system is the most common thermodynamic calculation method, and the research foundation is reliable high-temperature thermochemical data.
Prior art 1
In the current CVD experiment, the intermediate gas phase components are mainly measured by an in-situ Fourier infrared spectroscopy method, or important intermediate phases in the reaction process are determined by collecting CVD reaction tail gas and adopting a mass spectrum chromatographic analysis method, so that a deposition mechanism is established and explained, and the relationship between the deposition process conditions and the component ratio of the deposition products is qualitatively analyzed.
Disadvantages of the first prior art
Although the experiment has an intuitive observation result, the deposition usually occurs at a lower temperature and pressure due to the guarantee of the deposition quality, so that the deposition preparation period of the boron carbide is long and the production cost is high. In addition, the chemical vapor phase method has a special reaction system and harsh reaction conditions, so that the experimental determination is difficult to deeply and accurately determine.
Prior art 2
And (3) calculating and obtaining thermodynamic data of related gas-phase products in the multivariate system by utilizing a quantum chemistry combined statistical thermodynamic method. According to the chemical equilibrium principle, namely the chemical potential (Gibbs free energy) minimization principle, the equilibrium concentration distribution of relevant important products, especially solid-phase products under different process parameters (temperature, pressure and gas inlet ratio) is calculated. Through analysis and summarization, the optimal thermodynamic conditions of different solid phase product deposition are theoretically explained, the reaction rule is revealed, the preparation process parameters are optimized, and theoretical guidance is provided for experimental research.
The second prior art has the defects
When the range of the process conditions is determined, the composition problem of the deposition product can be well determined, but the single-objective or multi-objective optimization problem is difficult to solve, such as a process scheme for obtaining one or two highest deposition components cannot be found.
Abbreviation definitions
Error Back Propagation tracing (BP);
chemical Vapor Deposition (CVD);
genetic Algorithms (GA).
Definition of key terms
Ceramic matrix composite material: the ceramic matrix composite is a composite material compounded with various fibers by taking ceramic as a matrix;
chemical vapor deposition, which is a process of generating solid deposits by utilizing gaseous or steam substances to react on a gas phase or gas-solid interface;
numerical simulation, which is also called computer simulation. The purpose of researching engineering problems, physical problems and various problems in the nature is achieved by means of an electronic computer and by combining concepts of finite elements or finite volumes and through a method of numerical value calculation and image display;
machine learning is the science of artificial intelligence, and the main research object in the field is artificial intelligence, particularly how to use data and improve the performance of a specific algorithm in empirical learning.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a method for predicting the components of a chemical vapor multi-deposition product, which solves the defects in the prior art.
In order to realize the purpose, the technical scheme adopted by the invention is as follows:
a chemical vapor multi-deposition product composition prediction method comprises the following steps:
step S1, calculating a distribution function according to the process conditions:
translation partition function: when the molecular translation energy level difference is small, the translation partition function qtIs expressed as
Figure BDA0002689984210000031
Wherein m is the mass of the molecule, V is the volume of the molecule, T is the temperature, k is the boltzmann constant, and h is the planckian constant.
According to the formula of avogalois, pV-nrT-NAkT, where p is the gas pressure, R is the gas constant, NAIs an Avogastron constant, knowing that V is NAkT/p, obtained by substituting the formula (1)
Figure BDA0002689984210000041
Rotational distribution function: similarly, when the difference in rotational energy levels is small, the rotational partition function q of the moleculerThere is an analytical formula. For a linear molecule, the analytical formula is:
Figure BDA0002689984210000042
for nonlinear molecules then:
Figure BDA0002689984210000043
in the formula, I is the moment of inertia, and σ is a symmetric number.
Vibration partition function: due to the large difference in the vibration energy levels of the molecules, the vibration excitation follows statistical rules. Vibration partition function q of moleculeνAlso linear and nonlinear molecules. Taking the vibration ground state of the molecule as an energy zero point, the partition function of the linear molecule is as follows:
Figure BDA0002689984210000044
for nonlinear molecules then:
Figure BDA0002689984210000045
wherein n is the number of atoms contained in the molecule, (3n-5) or (3n-6) is the number of degrees of freedom of vibration or independent vibration modes, viIs the i-th in the moleculeThe vibration frequency (Hz) of the seed vibration mode.
Electronic allocation function: electronic distribution function qeThe expression of (a) is as follows:
Figure BDA0002689984210000051
in the formula, giIs the quantum weight or degeneracy of the ith electronic energy state,iis the energy of the ith state. The degree of degeneracy was calculated as follows: for a single atom, degree of degeneracy gi2J +1, wherein J is the quantum number of the total orbital angular momentum of the electrons; for polyatomic molecules, the degree of degeneracy is the product of the spin multiplicities and the irreducible representational dimensionality of the population of points to which each excited state belongs. Taking the degeneracy of the base state and the base state to be half of the total degeneracy of the base state.
Step S2, calculating the hot melt and entropy according to the translation, rotation, vibration and electronic distribution function
Total heat capacity (C) of gaseous monoatomic species taking into account the contributions of the translational, rotational, vibrational and electronic partition functions to heat capacity and entropyθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000052
Figure BDA0002689984210000053
total heat capacity (C) of gaseous diatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000054
Figure BDA0002689984210000055
wherein B is h/8c pi2I, c is the speed of light in vacuum, μ ═ ω/(kT), ω is the fundamental frequency of the harmonic oscillator.
Total heat capacity (C) of gaseous linear polyatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000061
Figure BDA0002689984210000062
total heat capacity (C) of gaseous nonlinear polyatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000063
Figure BDA0002689984210000064
step S3, calculating standard enthalpy and Gibbs free energy according to the translation, rotation, vibration and electronic distribution function;
standard enthalpy of formation ΔfHm θAnd standard Gibbs free energy of formationfGm θBy an atomization reaction [ formula (14)]And (4) calculating. The specific calculation formulas are shown in formula (15) and formula (16):
AmBnCxDy(gas)→mA(gas)+nB(gas)+xC(gas)+yD(gas) (14)
Figure BDA0002689984210000071
Figure BDA0002689984210000072
wherein, muiRepresents the stoichiometric number of the ith species, Δ of atom ifHm θ(i, g, T) and ΔfGm θ(i, g, T) is experimental data found in JANAF (or CODATA). In the formularHm θ(T) and. DELTA.rGm θ(T) is the reaction enthalpy change and the reaction Gibbs free energy change calculated from the following formulas:
Figure BDA0002689984210000073
Figure BDA0002689984210000074
wherein Hm θ(298.15K) is the standard enthalpy, C, at 298.15K calculated by the methods of G3(MP2) and G3// B3LYP combined with the electron energy obtained by statistical thermodynamic processingθ p,m(T) is the standard molar heat capacity fit, Sm θ(298.15K) is the standard entropy at 298.15K.
Step S4, obtaining an equilibrium yield distribution of all products from equation (19) according to the chemical equilibrium principle, i.e., the mathematical condition that the total gibbs free energy of the system is minimum:
Figure BDA0002689984210000075
wherein s is the total number of solid phase substances in the system, N is the total number of substances in the system, p is the total pressure, and N isiIs the amount of gas phase i species, ni cond.Are the masses of the solid phase i-th species, which satisfy the following relationship:
Figure BDA0002689984210000076
wherein, aijIs the atomic number of the element j in the species i, BjIs the total number of atoms of element j and M represents the total number of different elements.
Gas phase [ Delta G ]m θ(gas)]And solid phase [ Delta G ]m θ(cond.)]The standard molar Gibbs free energy of (a) is calculated by the Gibbs equation:
Figure BDA0002689984210000081
wherein the content of the first and second substances,
Figure BDA0002689984210000082
Figure BDA0002689984210000083
the heat capacity C in the above integralp,mThe heat capacity difference between the product of the formation reaction and the reactant is not present, and therefore the calculation is not enthalpy of formation or gibbs free energy of formation data.
And step S5, establishing a training model by using BP algorithm by taking the obtained various process conditions, deposition solid phase and yield as input data. Suppose there are N arbitrary samples (X)i,ti) Wherein X isi＝[xi1,xi2,…,xin]T∈Rn,ti＝[ti1,ti2,…,tim]T∈Rm. Wherein XiAs deposition conditions, tiThe molar number of each solid phase substance is calculated.
Figure BDA0002689984210000084
N arbitrary samples (X)i,ti) The method comprises the following specific steps of inputting the BP neural algorithm program: 1. network initialization, randomly giving each connection weight [ w ]],[v]And a threshold value thetai,rt(ii) a 2. Calculating hidden layers by a given input-output mode pair, and outputting the hidden layer units of the output layer; 3. calculating new connection weight and threshold; 4. and selecting the next input mode pair, returning to the second step, and repeatedly training until the network output error reaches the required training end.
And step S6, after a BP training model is established, the calculated result is analyzed by combining a genetic algorithm, and the highest solid phase yield under different weight conditions is continuously searched out through evolutionary operations such as selection, cross pairing, mutation and the like.
Compared with the prior art, the invention has the advantages that:
and the quantum chemical calculation, the thermodynamic calculation and the machine learning are combined in more detail to establish a model capable of accurately predicting the components of the deposition theory product. And establishing a relation between the obtained various process parameters and the components of the calculated deposition product by combining BP and GA algorithm, thereby realizing the optimization target of maximum multi-component yield. The optimization of the parameters of the deposition process can be carried out on the basis of the invention, and the CVD product with controllable components and good quality of the deposition product can be obtained. The invention can accelerate the development efficiency of CVD industrial production and reduce the production cost thereof, and provides a new idea and method for other CVD materials.
Drawings
FIG. 1 is a flow chart of a method for predicting the composition of a chemical vapor deposition product according to an embodiment of the present invention;
FIG. 2 is an exemplary diagram of a result of a balanced phase diagram calculation according to an embodiment of the present invention;
FIG. 3 is a graph showing the solid phase yield calculation results of the Si-B-C-N-H-Cl system according to the example of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings by way of examples.
As shown in fig. 1, a method for predicting the composition of a chemical vapor deposition product comprises the following steps:
step S1, calculating a distribution function according to the process conditions:
translation motionA distribution function: when the molecular translation energy level difference is small, the translation partition function qtIs expressed as
Figure BDA0002689984210000091
Wherein m is the mass of the molecule, V is the volume of the molecule, T is the temperature, k is the boltzmann constant, and h is the planckian constant.
According to the formula of avogalois, pV-nrT-NAkT (where p is the gas pressure, R is the gas constant, NAAvogalois constant), V ═ N is knownAkT/p, obtained by substituting the formula (1)
Figure BDA0002689984210000101
Rotational distribution function: similarly, when the difference in rotational energy levels is small, the rotational partition function q of the moleculerThere is an analytical formula. For a linear molecule, the analytical formula is:
Figure BDA0002689984210000102
for nonlinear molecules then:
Figure BDA0002689984210000103
in the formula, I is the moment of inertia, and σ is a symmetric number.
Vibration partition function: due to the large difference in the vibration energy levels of the molecules, the vibration excitation follows statistical rules. Vibration partition function q of moleculeνAlso linear and nonlinear molecules. Taking the vibration ground state (zero vibration energy level) of the molecule as the energy zero point, the partition function of the linear molecule is:
Figure BDA0002689984210000104
for nonlinear molecules then:
Figure BDA0002689984210000105
in the formula, n is the atom number contained in the molecule, (3n-5) or (3n-6) is the vibration freedom degree or independent vibration mode number (because the total freedom degree of the n atom molecule moving in three-dimensional space is 3n, wherein the translation freedom degree of the molecular mass center is 3, the linear molecule rotation is 2, the non-linear molecule rotation is 3, so 3n-5 or 3n-6 is the vibration freedom degree), viThe vibration frequency (Hz) of the ith vibration mode in the molecule.
Electronic allocation function: the electron excitation energy varies from molecule to molecule. At the higher temperatures typically used for CVD/CVI processes to produce materials, electron excitation is a non-negligible problem that must be taken into account in thermodynamic data calculations. Electronic distribution function qeThe expression of (a) is as follows:
Figure BDA0002689984210000111
in the formula, giIs the quantum weight (or degeneracy) of the ith electronic energy state,iis the energy of the ith state. The degree of degeneracy was calculated as follows: for a single atom, degree of degeneracy gi2J +1, J is the total (spin-orbit coupled) orbital angular momentum quantum number of electrons; for polyatomic molecules, the degree of degeneracy is the product of the spin multiplicities and the irreducible representational dimensionality of the population of points to which each excited state belongs. For individual molecules, due to some algorithmic drawbacks, a quasi-excited state is obtained that is very close to the ground state energy. The degeneracy of the ground state and the base state respectively accounts for half of the total degeneracy of the base state.
Step S2, calculating the hot melt and entropy according to the translation, rotation, vibration and electronic distribution function
Total heat capacity (C) of gaseous monoatomic species taking into account the contributions of the translational, rotational, vibrational and electronic partition functions to heat capacity and entropyθ p,m) Andentropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000112
Figure BDA0002689984210000113
total heat capacity (C) of gaseous diatomic molecules (ideal gases)θ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000114
Figure BDA0002689984210000121
wherein B is h/8c pi2I, c is the speed of light in vacuum, μ ═ ω/(kT), ω is the fundamental frequency of the harmonic oscillator.
Total heat capacity (C) of gaseous linear polyatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000122
Figure BDA0002689984210000123
total heat capacity (C) of gaseous nonlinear polyatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure BDA0002689984210000124
Figure BDA0002689984210000125
step S3, calculating standard enthalpy of formation and standard Gibbs free energy of formation according to translation, rotation, vibration and electronic distribution function
Standard enthalpy of formation ΔfHm θAnd standard Gibbs free energy of formationfGm θBy an atomization reaction [ formula (14)]And (4) calculating. The specific calculation formulas are shown in formula (15) and formula (16):
AmBnCxDy(gas)→mA(gas)+nB(gas)+xC(gas)+yD(gas)(14)
Figure BDA0002689984210000131
Figure BDA0002689984210000132
wherein, muiRepresents the stoichiometric number of the ith species, Δ of atom ifHm θ(i, g, T) and ΔfGm θ(i, g, T) is experimental data found in JANAF (or CODATA). In the formularHm θ(T) and. DELTA.rGm θ(T) is the reaction enthalpy change and the reaction Gibbs free energy change calculated from the following formulas:
Figure BDA0002689984210000133
Figure BDA0002689984210000134
wherein Hm θ(298.15K) is the electron energy calculated by the methods of G3(MP2) and G3// B3LYP combined with the standard enthalpy at 298.15K obtained from statistical thermodynamic processing,Cθ p,m(T) is the standard molar heat capacity fit, Sm θ(298.15K) is the standard entropy at 298.15K.
Step S4, the equilibrium yield distribution of all products can be obtained from formula (1) according to the chemical equilibrium principle, i.e. the mathematical condition that the total gibbs free energy (chemical potential) of the system is minimum:
Figure BDA0002689984210000135
wherein s is the total number of solid phase substances in the system, N is the total number of substances in the system, p is the total pressure, and N isiIs the amount of gas phase i species, ni cond.Are the masses of the solid phase i-th species, which satisfy the following relationship:
Figure BDA0002689984210000136
wherein, aijIs the atomic number of the element j in the species i, BjIs the total number of atoms of element j and M represents the total number of different elements.
Gas phase [ Delta G ]m θ(gas)]And solid phase [ Delta G ]m θ(cond.)]Is calculated from the Gibbs equation:
Figure BDA0002689984210000141
wherein the content of the first and second substances,
Figure BDA0002689984210000142
Figure BDA0002689984210000143
the heat capacity C in the above integralp,mThe heat capacity difference between the product of the formation reaction and the reactant is not present, and therefore the calculation is not enthalpy of formation or gibbs free energy of formation data. The partial calculation results are shown in fig. 2 and 3.
And step S5, establishing a training model by using BP algorithm by taking the obtained various process conditions, deposition solid phase and yield as input data. Suppose there are N arbitrary samples (X)i,ti) Wherein X isi＝[xi1,xi2,…,xin]T∈Rn,ti＝[ti1,ti2,…,tim]T∈Rm. Wherein XiAs deposition conditions, tiThe molar number of each solid phase substance is calculated.
Figure BDA0002689984210000144
The BP algorithm consists of two processes, forward computation (forward propagation) of the data stream and back propagation of the error signal. In forward propagation, the propagation direction is input layer → hidden layer → output layer, and the state of each layer of neurons only affects the next layer of neurons. If the desired output is not available at the input layer, the back propagation flow of the error signal is reversed. By alternately carrying out the two processes, an error function gradient descending strategy is executed in the weight vector space, and a group of weight vectors are dynamically searched to ensure that the network error function reaches the minimum value. N arbitrary samples (X)i,ti) The method comprises the following specific steps of inputting the BP neural algorithm program: 1. network initialization, randomly giving each connection weight [ w ]],[v]And a threshold value thetai,rt(ii) a 2. Calculating hidden layers by a given input-output mode pair, and outputting the hidden layer units of the output layer; 3. calculating new connection weight and threshold; 4. and selecting the next input mode pair, returning to the second step, and repeatedly training until the network output error reaches the required training end.
And step S6, after a BP training model is established, the calculated result is analyzed by combining a genetic algorithm, and the highest solid phase yield under different weight conditions is continuously searched out through evolutionary operations such as selection, cross pairing, mutation and the like.
It will be appreciated by those of ordinary skill in the art that the examples described herein are intended to assist the reader in understanding the manner in which the invention is practiced, and it is to be understood that the scope of the invention is not limited to such specifically recited statements and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.

Claims (1)

1. A method for predicting the components of a chemical vapor multi-deposition product is characterized by comprising the following steps:
step S1, calculating a distribution function according to the process conditions:
translation partition function: when the molecular translation energy level difference is small, the translation partition function qtIs expressed as
Figure FDA0002689984200000011
Wherein m is the mass of the molecule, V is the volume of the molecule, T is the temperature, k is the Boltzmann constant, and h is the Planckian constant;
according to the formula of avogalois, pV-nrT-NAkT, where p is the gas pressure, R is the gas constant, NAIs an Avogastron constant, knowing that V is NAkT/p, obtained by substituting the formula (1)
Figure FDA0002689984200000012
Rotational distribution function: similarly, when the difference in rotational energy levels is small, the rotational partition function q of the moleculerHas an analytic formula; for a linear molecule, the analytical formula is:
Figure FDA0002689984200000013
for nonlinear molecules then:
Figure FDA0002689984200000014
in the formula, I is rotational inertia, and sigma is a symmetric number;
vibration partition function: the vibration excitation follows a statistical rule because the vibration energy level difference of molecules is large; vibration partition function q of moleculeνLinear and nonlinear molecules are also distinguished; taking the vibration ground state of the molecule as an energy zero point, the partition function of the linear molecule is as follows:
Figure FDA0002689984200000015
for nonlinear molecules then:
Figure FDA0002689984200000021
wherein n is the number of atoms contained in the molecule, (3n-5) or (3n-6) is the number of degrees of freedom of vibration or independent vibration modes, viThe vibration frequency (Hz) of the ith vibration mode in the molecule;
electronic allocation function: electronic distribution function qeThe expression of (a) is as follows:
Figure FDA0002689984200000022
in the formula, giIs the quantum weight or degeneracy of the ith electronic energy state,iis the energy of the ith state; the degree of degeneracy was calculated as follows: for a single atom, degree of degeneracy gi2J +1, wherein J is the quantum number of the total orbital angular momentum of the electrons; for polyatomic molecules, the degree of degeneracy is the product of the spin multiplicities and the irreducible dimensionality of the point group to which each excited state belongs; taking the degree of degeneracy of it with the ground stateEach account for half of the total degeneracy of the ground state;
step S2, calculating the hot melt and entropy according to the translation, rotation, vibration and electronic distribution function
Total heat capacity (C) of gaseous monoatomic species taking into account the contributions of the translational, rotational, vibrational and electronic partition functions to heat capacity and entropyθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure FDA0002689984200000023
Figure FDA0002689984200000024
total heat capacity (C) of gaseous diatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure FDA0002689984200000031
Figure FDA0002689984200000032
Figure FDA0002689984200000033
wherein B is h/8c pi2I, c is the speed of light in vacuum, μ ═ ω/(kT), ω is the fundamental frequency of the harmonic oscillator;
total heat capacity (C) of gaseous linear polyatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure FDA0002689984200000034
Figure FDA0002689984200000035
total heat capacity (C) of gaseous nonlinear polyatomic moleculesθ p,m) And entropy (S)θ m) The expression of (a) is:
Figure FDA0002689984200000036
Figure FDA0002689984200000041
step S3, calculating standard enthalpy and Gibbs free energy according to the translation, rotation, vibration and electronic distribution function;
standard enthalpy of formation ΔfHm θAnd standard Gibbs free energy of formationfGm θBy an atomization reaction [ formula (14)]Calculating; the specific calculation formulas are shown in formula (15) and formula (16):
AmBnCxDy(gas)→mA(gas)+nB(gas)+xC(gas)+yD(gas) (14)
Figure FDA0002689984200000042
Figure FDA0002689984200000043
wherein, muiRepresents the stoichiometric number of the ith species, Δ of atom ifHm θ(i, g, T) and ΔfGm θ(i, g, T) are experimental data found in JANAF (or CODATA); in the formularHm θ(T) and. DELTA.rGm θ(T) is the reaction enthalpy change and the reaction Gibbs free energy change calculated from the following formulas:
Figure FDA0002689984200000044
Figure FDA0002689984200000045
wherein Hm θ(298.15K) is the standard enthalpy, C, at 298.15K calculated by the methods of G3(MP2) and G3// B3LYP combined with the electron energy obtained by statistical thermodynamic processingθ p,m(T) is the standard molar heat capacity fit, Sm θ(298.15K) is standard entropy at 298.15K;
step S4, obtaining an equilibrium yield distribution of all products from equation (19) according to the chemical equilibrium principle, i.e., the mathematical condition that the total gibbs free energy of the system is minimum:
Figure FDA0002689984200000051
wherein s is the total number of solid phase substances in the system, N is the total number of substances in the system, p is the total pressure, and N isiIs the amount of gas phase i species, ni cond.Are the masses of the solid phase i-th species, which satisfy the following relationship:
Figure FDA0002689984200000052
wherein, aijIs the atomic number of the element j in the species i, BjIs the total atomic number of element j, and M represents the total number of different elements;
gas phase [ Delta G ]m θ(gas)]And solid phase [ Delta G ]m θ(cond.)]The standard molar Gibbs free energy of (a) is calculated by the Gibbs equation:
Figure FDA0002689984200000053
wherein the content of the first and second substances,
Figure FDA0002689984200000054
Figure FDA0002689984200000055
the heat capacity C in the above integralp,mThe heat capacity difference between the product of the forming reaction and the reactant is not, and therefore, the calculation result is not enthalpy of formation or gibbs free energy data;
step S5, establishing a training model by using BP algorithm by taking the obtained various process conditions, deposition solid phase and yield as input data; suppose there are N arbitrary samples (X)i,ti) Wherein X isi＝[xi1,xi2,…,xin]T∈Rn,ti＝[ti1,ti2,…,tim]T∈Rm(ii) a Wherein XiAs deposition conditions, tiThe calculated mole number of each solid phase substance;
Figure FDA0002689984200000061
n arbitrary samples (X)i,ti) The method comprises the following specific steps of inputting the BP neural algorithm program: 1. network initialization, randomly giving each connection weight [ w ]],[v]And a threshold value thetai,rt(ii) a 2. Calculating hidden layers by a given input-output mode pair, and outputting the hidden layer units of the output layer; 3. calculating new connection weight and threshold; 4. selecting the next input mode pair, returning to the second step, and repeatedly training until the network output error reaches the required training end;
and step S6, after a BP training model is established, the calculated result is analyzed by combining a genetic algorithm, and the highest solid phase yield under different weight conditions is continuously searched out through evolutionary operations such as selection, cross pairing, mutation and the like.
CN202010988337.3A 2020-09-18 2020-09-18 Chemical vapor multicomponent deposition product component prediction method Active CN112102897B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010988337.3A CN112102897B (en) 2020-09-18 2020-09-18 Chemical vapor multicomponent deposition product component prediction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010988337.3A CN112102897B (en) 2020-09-18 2020-09-18 Chemical vapor multicomponent deposition product component prediction method

Publications (2)

Publication Number Publication Date
CN112102897A true CN112102897A (en) 2020-12-18
CN112102897B CN112102897B (en) 2022-03-08

Family

ID=73759627

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010988337.3A Active CN112102897B (en) 2020-09-18 2020-09-18 Chemical vapor multicomponent deposition product component prediction method

Country Status (1)

Country Link
CN (1) CN112102897B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112687347A (en) * 2021-01-05 2021-04-20 西安交通大学 Method and system for calculating arc plasma components under chemical non-equilibrium effect
CN113160898A (en) * 2021-05-18 2021-07-23 北京信息科技大学 Prediction method and system for Gibbs free energy of iron-based alloy
CN114510820A (en) * 2021-12-31 2022-05-17 西安交通大学 Coal thermodynamic parameter construction method
CN114943195A (en) * 2022-06-02 2022-08-26 西南石油大学 Method for constructing water gas shift reaction equilibrium constant prediction model
CN114944206A (en) * 2022-05-20 2022-08-26 苏州大学 Method for predicting phase stability and phase fraction of 3D printing multi-principal element alloy

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0178200A1 (en) * 1984-09-10 1986-04-16 FAIRCHILD CAMERA & INSTRUMENT CORPORATION Method of control for chemical vapor deposition
US20090243105A1 (en) * 2008-03-31 2009-10-01 Matthias Lehr Wire bonding on reactive metal surfaces of a metallization of a semiconductor device by providing a protective layer
US20120084238A1 (en) * 2007-07-31 2012-04-05 Cornell Research Foundation, Inc. System and Method to Enable Training a Machine Learning Network in the Presence of Weak or Absent Training Exemplars
CN110021372A (en) * 2017-07-13 2019-07-16 中国石油化工股份有限公司 A kind of system for predicting multicomponent system vapor-liquid equilibrium
CN110598255A (en) * 2019-08-14 2019-12-20 华南理工大学 Chemical vapor deposition rate prediction method
CN111597735A (en) * 2020-06-19 2020-08-28 华南理工大学 Component prediction method combining machine learning and CVD modeling

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0178200A1 (en) * 1984-09-10 1986-04-16 FAIRCHILD CAMERA & INSTRUMENT CORPORATION Method of control for chemical vapor deposition
US20120084238A1 (en) * 2007-07-31 2012-04-05 Cornell Research Foundation, Inc. System and Method to Enable Training a Machine Learning Network in the Presence of Weak or Absent Training Exemplars
US20090243105A1 (en) * 2008-03-31 2009-10-01 Matthias Lehr Wire bonding on reactive metal surfaces of a metallization of a semiconductor device by providing a protective layer
CN110021372A (en) * 2017-07-13 2019-07-16 中国石油化工股份有限公司 A kind of system for predicting multicomponent system vapor-liquid equilibrium
CN110598255A (en) * 2019-08-14 2019-12-20 华南理工大学 Chemical vapor deposition rate prediction method
CN111597735A (en) * 2020-06-19 2020-08-28 华南理工大学 Component prediction method combining machine learning and CVD modeling

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
关康,任海涛等: "使用新的微观结构参数估计多孔陶瓷的热导率和弹性模量", 《欧洲陶瓷学会杂志》 *
邓腾飞，卢振亚等: "化学计量比偏移对钛酸铜钙陶瓷微观结构和电性能的影响", 《稀有金属材料与工程》 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112687347A (en) * 2021-01-05 2021-04-20 西安交通大学 Method and system for calculating arc plasma components under chemical non-equilibrium effect
CN112687347B (en) * 2021-01-05 2024-04-02 西安交通大学 Method and system for calculating arc plasma component under chemical unbalance effect
CN113160898A (en) * 2021-05-18 2021-07-23 北京信息科技大学 Prediction method and system for Gibbs free energy of iron-based alloy
CN113160898B (en) * 2021-05-18 2023-09-08 北京信息科技大学 Iron-based alloy Gibbs free energy prediction method and system
CN114510820A (en) * 2021-12-31 2022-05-17 西安交通大学 Coal thermodynamic parameter construction method
CN114510820B (en) * 2021-12-31 2024-05-07 西安交通大学 Coal thermodynamic parameter construction method
CN114944206A (en) * 2022-05-20 2022-08-26 苏州大学 Method for predicting phase stability and phase fraction of 3D printing multi-principal element alloy
CN114944206B (en) * 2022-05-20 2024-03-19 苏州大学 Method for predicting phase stability and phase fraction of 3D printing multi-principal-element alloy
CN114943195A (en) * 2022-06-02 2022-08-26 西南石油大学 Method for constructing water gas shift reaction equilibrium constant prediction model
CN114943195B (en) * 2022-06-02 2024-04-26 西南石油大学 Construction method of water gas shift reaction equilibrium constant prediction model

Also Published As

Publication number Publication date
CN112102897B (en) 2022-03-08

Similar Documents

Publication Publication Date Title
CN112102897B (en) Chemical vapor multicomponent deposition product component prediction method
D'angola et al. Thermodynamic and transport properties in equilibrium air plasmas in a wide pressure and temperature range
CN111597735B (en) Component prediction method combining machine learning and CVD modeling
Shigeta et al. Growth model of binary alloy nanopowders for thermal plasma synthesis
Biegler Advances in nonlinear programming concepts for process control
Te Braake et al. Semi-mechanistic modeling of chemical processes with neural networks
Birakayala et al. A reduced reaction model for carbon CVD/CVI processes
Scott Day et al. Classical trajectory study of collisions of Ar with alkanethiolate self-assembled monolayers: Potential-energy surface effects on dynamics
CN110045606A (en) A kind of increment space-time learning method for distributed parameter system line modeling
Kobayashi et al. Self-learning hybrid Monte Carlo method for isothermal–isobaric ensemble: Application to liquid silica
Wang et al. Energy transfer in intermolecular collisions of polycyclic aromatic hydrocarbons with bath gases He and Ar
Griesemer et al. Accelerating the prediction of stable materials with machine learning
Han et al. Predicting mechanical properties of ultrahigh temperature ceramics using machine learning
CN112331271B (en) Cross-scale prediction method for CVD deposition rate, product texture and quality
Baeurle et al. Grand canonical investigations of prototypical polyelectrolyte models beyond the mean field level of approximation
Wang et al. Prediction of Gibbs free energy for hydrogen
Mien Design of IMC-PI controller for nonlinear model of continuous stirred tank reactor
Schnack et al. A coupled numerical modelling and experimental approach in chemical vapour infiltration (CVI) process of SiC/SiC composites
CN114943195B (en) Construction method of water gas shift reaction equilibrium constant prediction model
Fokin et al. Joint Russian and Bulgarian academies of sciences database of intermolecular potentials and diffusion coefficients for components of the CVD processes in microelectronics
Döntgen Analytic energy-level densities of separable harmonic oscillators including approximate hindered rotor corrections
CN112687345B (en) Prediction method and prediction device for phosphoric acid extraction process
Łach et al. Recent Progress in Heat and Mass Transfer Modeling for Chemical Vapor Deposition Processes
Jones Intelligent process control of fiber chemical vapor deposition
Alrasheed Effective Optimization Strategies for Abrasive Water Jet Machining of Glass-Carbon Fiber Reinforced Composites: A Comparative Study of Evolutionary Optimization Techniques

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right

Effective date of registration: 20240202

Address after: Room 124, 1st to 2nd floors, No.124 Xingsheng Street, Huangcun Town, Daxing District, Beijing, 102627 (cluster registration)

Patentee after: Beijing Huifengyuan Technology Co.,Ltd.

Country or region after: China

Address before: 510641 No. five, 381 mountain road, Guangzhou, Guangdong, Tianhe District

Patentee before: SOUTH CHINA University OF TECHNOLOGY

Country or region before: China

TR01 Transfer of patent right